Electric heating collaboration method for directional transportation of multi-scale electric heat energy flow

Abstract

The invention discloses an electric heating cooperation method for multi-scale electric heat energy flow directional transportation, and belongs to the technical field of electric power, and the method comprises the steps: obtaining the temperature, generating capacity, voltage and current data of a heat storage system and a photovoltaic system in real time, predicting the photovoltaic generating capacity and the heat storage generating capacity through a regression analysis and decision-making tree model, and carrying out the prediction of the photovoltaic generating capacity and the heat storage generating capacity according to the historical electricity consumption data. On the basis of a control decision, optimization adjustment is carried out, user loads are predicted through an ARIMA model, the technical problem of non-coordination of the heat storage system and the photovoltaic system in the energy utilization and scheduling process is solved, efficient cooperation of the heat storage system and the photovoltaic system is achieved, heat storage and heat release modes are dynamically adjusted according to historical and real-time data, and the energy utilization and scheduling efficiency is improved. A machine learning model is adopted to improve the accuracy of generating capacity and load prediction, and the response capability of the system to environment change is enhanced.

Description

Technical Field

[0001] The present invention belongs to the field of electric power technology, and in particular relates to an electrothermal synergistic method for directional transport of multi-scale electrothermal energy flows. Background Art

[0002] With the rapid development of renewable energy technology, the combination of photovoltaic power generation and thermal storage system has attracted increasing attention. This system converts solar energy into electrical energy, stores thermal energy in the thermal storage system, and supplements the insufficient power generation of the photovoltaic system by releasing heat and storing heat, thereby achieving the coordinated use of electricity and thermal energy and improving energy utilization efficiency.

[0003] In this system, the photovoltaic system is responsible for converting solar radiation into electrical energy. The system monitors voltage and current in real time to calculate the power generation. Through an integrated control system, these data are combined with the thermal storage temperature and power generation data of the thermal storage system. The thermal storage system stores thermal energy when there is excess power generation and releases heat during peak demand to heat buildings or generate electricity.

[0004] The existing technology has some defects: Energy management is not smart enough: Many existing systems still rely on simple rules and thresholds for energy scheduling and management, and lack real-time monitoring and dynamic adjustment capabilities. This results in insufficient flexibility in system response and inability to effectively cope with photovoltaic power generation fluctuations and load changes; Insufficient prediction accuracy: Current prediction models often rely on historical data and fail to fully consider a variety of influencing factors (such as weather changes), resulting in inaccurate prediction results and affecting subsequent decision-making. Summary of the invention

[0005] The purpose of the present invention is to provide an electrothermal coordination method for directional transport of multi-scale electrothermal energy flows, which solves the technical problem of solving the incoordination between the thermal storage system and the photovoltaic system in the process of energy utilization and scheduling.

[0006] To achieve the above object, the present invention adopts the following technical solution: An electrothermal synergistic method for directional transport of multi-scale electrothermal energy flow comprises the following steps: Step 1: The data collection module obtains the water storage temperature data Tstorage, the outlet water temperature data Tout and the return water temperature data Treturn of the thermal storage system in real time through the Internet, constructs a thermal storage temperature data set, and simultaneously obtains the voltage Vthermal and current Ithermal of the thermal storage system power generation output, calculates the power generation Pthermal of the thermal storage system; constructs a thermal storage power generation data set; The data collection module obtains the voltage Vpv and current Ipv output by the photovoltaic system in real time through the Internet, calculates the power generation Ppv of the photovoltaic system, and generates a photovoltaic power data set; Step 2: The prediction module retrieves the photovoltaic power dataset and predicts the photovoltaic power generation Ppv,forecast of the photovoltaic system in the next time period t through regression analysis; Retrieve the thermal energy storage power generation dataset and predict the thermal energy storage power generation Pthermal,forecast in the next time period t through a decision tree regression model; Step 3: The prediction module calculates the total predicted power generation Ptotal,forecast; Ptotal,forecast = Ppv,forecast + Pthermal,forecast; Step 4: The control decision module obtains the historical heating demand data and historical power consumption data. According to the historical power consumption data, draw a time series graph of historical power consumption, and calculate the predicted user load Pload,next in the next time period t according to the ARIMA model; Retrieve the total predicted power generation Ptotal,forecast calculated by the prediction module, make a decision, and optimize it: Decision: If Ptotal,forecast < Pload,next, the thermal energy storage system enters the heat release mode; If Ptotal,forecast ≥ Pload,next + s, the thermal energy storage system enters the heat storage mode; Optimized decision: If Tout < Tdemand, the thermal energy storage system gives priority to heating; If Tstorage > Tmax, the thermal energy storage system considers entering the heat release mode; Tdemand is the target temperature of the user's heating demand; Tmax is the maximum safe temperature of the water stored in the thermal energy storage system; s is the preset safety margin; According to the load fluctuation frequency Fload, the power generation change frequency Fpv, the system response time Tresponse, and the heating demand change Fheat, calculate the duration Tswitch of the mode switching of the thermal energy storage system according to the following formula: T switch = k × [ 1 / (F load + F pv + F heat )] + T response ; Among them, k is the adjustment coefficient, which is adjusted according to the actual operation of the thermal storage system and reflects the influence of various factors on the duration; Step 5: The control decision module sends the control decision to the thermal storage system. The thermal storage system starts the corresponding valves and pumps according to the control decision, adjusts the coordinated exchange of thermal energy and electrical energy, and realizes dynamic regulation of energy.

[0007] Preferably, when executing step 1, the power generation Pthermal of the thermal storage system is calculated by the following formula: Pthermal = Vthermal × Ithermal; The power generation Ppv of the photovoltaic system is calculated by the following formula: Ppv = Vpv × Ipv.

[0008] Preferably, when executing step 2, the specific calculation formula of photovoltaic power generation Ppv,forecast is as follows: P pv,forecast = a 0 + a 1 × P pv,real + a 2 × P load,hist + a 3 × Weather Factors + ϵ ; Among them, Ppv,forecas is the predicted photovoltaic power generation; a0 is the intercept, which represents the power generation when all independent variables are 0; a1, a2 and a3 are regression coefficients, which respectively represent the influence of real-time photovoltaic power generation, historical power consumption and weather factors on power generation; ϵ is the error term, which represents the prediction error of the model; Ppv,real is the real-time photovoltaic system power generation, Ppv,hist is the historical photovoltaic power generation data, and Ppv,real and Ppv,hist are calculated using the data in the photovoltaic power data set; Weather Factors is weather data, including solar radiation intensity; The specific calculation formula of thermal storage power generation Pthermal,forecast is as follows: For a given input feature vector X, use the trained decision tree model to make predictions: X = [Tstorage,Tout,Treturn,Pthermal,past]; Pthermal,forecast = Tree(X); Pthermal,past refers to the power generation data of the thermal storage system in a certain period of time in the past, which is obtained through the thermal storage power generation data set; Tree represents the decision tree model.

[0009] Preferably, when executing step 4, the historical electricity consumption data is the historical user electricity consumption data obtained by the control decision module from the power grid system via the Internet; The execution steps of the control decision module are as follows: Step 4-1: The control decision module obtains historical power consumption data; Step 4-2: The control decision module cleans the historical power consumption data, removes outliers and missing values, constructs a historical power consumption data set, and arranges the data in the historical power consumption data set in time series, specifically in hourly, daily or weekly time series; Step 4-3: Draw a time series graph to obtain the trend, seasonality and periodicity characteristics of the data; draw the autocorrelation graph ACF and partial autocorrelation graph PACF; Step 4-4: Determine the parameters of the ARIMA model, including the number of autoregressive terms p, the number of differences d, and the number of moving average terms q. The number of autoregressive terms p is the number of significant lags obtained from the PACF diagram, the number of moving average terms q is the number of significant lags obtained from the ACF diagram, and the number of differences d is the number of differences performed when the unit root test is used to check the stationarity of the time series. Step 4-5: Build ARIMA model: ARIMA(p,d,q); Step 4-6: Fit the ARIMA model using historical electricity consumption data; Step 4-7: Use the fitted ARIMA model to predict the next time period. The prediction formula is: P ＇ load,t = μ + ϕ 1 P load,t−1 + ϕ 2 P load,t−2 + ...+ ϕ p1 P load,t−p1 + i 1 ϵ t−1 + i 2 ϵt−2 +...+ i q ϵ t−q ; in, P ＇ load,t is the predicted user load; μ is the constant term of the model; ϕ is the autoregressive coefficient; θ is the moving average coefficient; ϵ is the error term; Step 4-8: Use the predicted value as the predicted user load Pload,next for the next time period; Step 4-9: Use the mean square error (MSE) or root mean square error (RMSE) indicator to evaluate the predictive ability of the ARIMA model; Step 4-10: Regularly adjust the ARIMA model parameters based on the difference between actual electricity consumption and predicted values.

[0010] Preferably, the load fluctuation frequency Fload is obtained by analyzing the changes in historical power consumption data and calculating the frequency of load changes per unit time; The power generation change frequency Fpv is obtained by monitoring the power generation of the photovoltaic system and recording the number of changes in power generation per unit time; The system response time Tresponse is determined by the system's reaction time to changes in the input signal and is a preset fixed value; The heating demand change Fheat is obtained by analyzing the changes in historical heating demand data, which represents the number of heating demand changes within a statistical unit time.

[0011] The electric-thermal coordination method for directional transport of multi-scale electric-thermal energy flows described in the present invention solves the technical problem of solving the incoordination between the heat storage system and the photovoltaic system in the process of energy utilization and scheduling. The present invention realizes efficient coordination between the heat storage system and the photovoltaic system, can monitor and predict the demand for electric energy and thermal energy in real time, dynamically adjust the heat storage and release modes according to historical and real-time data, thereby optimizing energy utilization efficiency, and adopts machine learning models to improve the accuracy of power generation and load prediction, and enhance the system's responsiveness to environmental changes. It not only reduces energy waste, but also effectively improves the utilization rate of renewable energy, thereby facilitating intelligent energy management. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is the main flow chart of the present invention; Figure 2 It is a flow chart of step 4 of the present invention. DETAILED DESCRIPTION

[0013] Depend on Figure 1-Figure 2 The electrothermal synergistic method for directional transport of multi-scale electrothermal energy flow shown in the figure comprises the following steps: Step 1: The data collection module obtains the water storage temperature data Tstorage, the outlet water temperature data Tout and the return water temperature data Treturn of the thermal storage system in real time through the Internet, constructs a thermal storage temperature data set, and simultaneously obtains the voltage Vthermal and current Ithermal of the thermal storage system power generation output, calculates the power generation Pthermal of the thermal storage system; constructs a thermal storage power generation data set; The power generation Pthermal of the thermal storage system is calculated by the following formula: Pthermal = Vthermal × Ithermal; The data collection module obtains the voltage Vpv and current Ipv output by the photovoltaic system in real time through the Internet, calculates the power generation Ppv of the photovoltaic system, and generates a photovoltaic power data set; The power generation Ppv of the photovoltaic system is calculated by the following formula: Ppv = Vpv × Ipv.

[0014] In this embodiment, the frequency of data acquisition is once a minute to ensure real-time performance.

[0015] Step 2: The prediction module retrieves the photovoltaic power data set and predicts the photovoltaic power generation Ppv,forecast of the photovoltaic system in the next time period t through regression analysis; The specific calculation formula of photovoltaic power generation Ppv,forecast is as follows: P pv,forecast = a 0 + a 1 × P pv,real + a 2 × P load,hist + a 3 × Weather Factors + ϵ ; Among them, Ppv,forecas is the predicted photovoltaic power generation; a0 is the intercept, which represents the power generation when all independent variables are 0; a1, a2 and a3 are regression coefficients, which respectively represent the influence of real-time photovoltaic power generation, historical power consumption and weather factors on power generation; ϵ is the error term, which represents the prediction error of the model; Ppv,real is the real-time photovoltaic system power generation, Ppv,hist is the historical photovoltaic power generation data, and Ppv,real and Ppv,hist are calculated using the data in the photovoltaic power data set; Weather Factors refers to weather data, including solar radiation intensity. Solar radiation intensity can usually be obtained through meteorological data services.

[0016] Retrieve the thermal storage power generation data set and use the decision tree regression model to predict the thermal storage power generation Pthermal,forecast in the next time period t; The specific calculation formula of thermal storage power generation Pthermal,forecast is as follows: For a given input feature vector X, use the trained decision tree model to make predictions: X = [Tstorage,Tout,Treturn,Pthermal,past]; Pthermal,forecast = Tree(X); Pthermal,past refers to the power generation data of the thermal storage system in a certain period of time in the past, which is obtained through the thermal storage power generation data set; Tree represents the decision tree model.

[0017] In the regression analysis of this embodiment, a multivariate linear regression or random forest regression model is used to improve the prediction accuracy. The effect of each model will be evaluated by the mean square error (MSE) to ensure the reliability of the prediction results.

[0018] Step 3: The forecast module calculates the total forecast power generation Ptotal,forecast; Ptotal,forecast = Ppv,forecast + Pthermal,forecast; Step 4: The control decision module obtains historical heating demand data and historical electricity consumption data, draws a time series diagram of historical electricity consumption based on the historical electricity consumption data, and calculates the predicted user load Pload,next for the next time period t based on the ARIMA model; The historical electricity consumption data is the historical user electricity consumption data obtained by the control decision module from the power grid system through the Internet; The execution steps of the control decision module are as follows: Step 4-1: The control decision module obtains historical power consumption data; collects and organizes historical power consumption data Pload,t, and ensures that the data is arranged in time series (for example, by hour, day or week). The data should include multiple time records for easy analysis.

[0019] Step 4-2: The control decision module cleans the historical power consumption data, removes outliers and missing values, constructs a historical power consumption data set, and arranges the data in the historical power consumption data set in time series, specifically in hourly, daily or weekly time series; In this embodiment, during the data cleaning process, the Z-score or IQR method can be used to detect outliers to ensure data quality.

[0020] Step 4-3: Draw a time series graph to obtain the trend, seasonality and periodicity characteristics of the data; draw the autocorrelation graph ACF and partial autocorrelation graph PACF; Draw a time series graph of historical electricity consumption to observe trends and seasonal changes in the data. Determine whether there is a clear upward or downward trend and whether there are cyclical fluctuations.

[0021] The autocorrelation plot ACF and partial autocorrelation plot PACF are used to help determine the autocorrelation of the data and select the parameters of the ARIMA model.

[0022] Step 4-4: Determine the parameters of the ARIMA model, including the number of autoregressive terms p, the number of differences d, and the number of moving average terms q. The number of autoregressive terms p is the number of significant lags obtained from the PACF diagram, the number of moving average terms q is the number of significant lags obtained from the ACF diagram, and the number of differences d is the number of differences performed when the unit root test is used to check the stationarity of the time series.

[0023] Use a unit root test (such as the Augmented Dickey-Fuller test) to check the stationarity of the time series. The mean and variance of a stationary series do not change over time. If the series is not stationary, then it needs to be differencing: Pload,diff = Pload,t − Pload,t−1; Perform the difference until the series is stationary, and record the number of differences d.

[0024] Step 4-5: Build ARIMA model: ARIMA(p,d,q); Step 4-6: Fit the ARIMA model using historical electricity consumption data; the fitting process includes estimating the model parameters using the maximum likelihood estimation method (MLE).

[0025] Step 4-7: Use the fitted ARIMA model to predict the next time period. The prediction formula is: P ＇ load,t = μ + ϕ 1 P load,t−1 + ϕ 2 P load,t−2 + ...+ ϕ p1P load,t−p1 + i 1 ϵ t−1 + i 2 ϵ t−2 +...+ i q ϵ t−q ; wherein, P ＇ load,t is the predicted user load; μ is the constant term of the model; ϕ is the autoregressive coefficient; θ is the moving average coefficient; ϵ is the error term; t represents the time period; q is the number of moving average terms; p1 is the number of autoregressive terms; load represents the user load side.

[0026] Steps 4-8: Take the predicted value P ＇ load,t as the predicted user load Pload,next for the next time period; Step 4-9: Evaluate the prediction ability of the ARIMA model using the mean squared error MSE or root mean squared error RMSE metric; Step 4-10: Regularly adjust the ARIMA model parameters according to the difference between the actual power consumption and the predicted value.

[0027] Retrieve the total predicted power generation Ptotal,forecast calculated by the prediction module, make decisions, and optimize: Decision: If Ptotal,forecast < Pload,next, the heat storage system enters the heat release mode; If Ptotal,forecast ≥ Pload,next + s, the heat storage system enters the heat storage mode; Optimized decision: If Tout < Tdemand, the heat storage system gives priority to heat supply; If Tstorage > Tmax, the heat storage system considers entering the heat release mode; Tdemand is the target temperature of the user's heat supply demand; Tmax is the maximum safe temperature of the stored water in the heat storage system; s is the preset safety margin; In this embodiment, the value of s is the percentage value of Pload,next, and its value can consider the following factors: Load volatility: If the user's load varies greatly, it is recommended to set a larger safety margin to ensure that the demand can be met during peak hours. Generally, the value of s can be considered to be between 5% and 15%, depending on the fluctuation range of historical load data.

[0028] Thermal storage system response time: If the thermal storage system has a long response time, it is recommended to set a higher safety margin to ensure that the system can adjust in time in the event of a sudden increase in load. It can be considered to be between 10% and 20%.

[0029] Thermal storage system efficiency: If the thermal storage system has a higher efficiency, the safety margin can be reduced appropriately because the system can adapt to load changes more quickly. In this case, s can be set to 5% to 10%.

[0030] Taking all factors into consideration when setting the value of s, a reasonable initial value can generally be between 5% and 15%.

[0031] According to the load fluctuation frequency Fload, the power generation change frequency Fpv, the system response time Tresponse and the heating demand change Fheat, the mode switching duration Tswitch of the thermal storage system is calculated according to the following formula: T switch = k × [ 1 / (F load + F pv + F heat )] + T response ; Among them, k is the adjustment coefficient, which is adjusted according to the actual operation of the thermal storage system and reflects the influence of various factors on the duration; The load fluctuation frequency Fload is obtained by analyzing the changes in historical power consumption data and calculating the frequency of load changes per unit time; The power generation change frequency Fpv is obtained by monitoring the power generation of the photovoltaic system and recording the number of changes in power generation per unit time; The system response time Tresponse is determined by the system's reaction time to changes in the input signal and is a preset fixed value; The heating demand change Fheat is obtained by analyzing the changes in historical heating demand data, which represents the number of heating demand changes within a statistical unit time.

[0032] Step 5: The control decision module sends the control decision to the thermal storage system. The thermal storage system starts the corresponding valves and pumps according to the control decision, adjusts the coordinated exchange of thermal energy and electrical energy, and realizes dynamic regulation of energy.

[0033] The control system of valves and pumps can use PID control algorithm to achieve precise adjustment. At the same time, the control system will monitor the working status of valves and pumps in real time to ensure the safety and efficiency of the system.

[0034] In this embodiment, the valve is used to regulate the flow of fluid, control the flow direction of hot water in the thermal storage system, and ensure the effective storage and release of thermal energy. When the system enters the heat release mode, the valve will open to allow hot water to flow out of the thermal storage tank to meet the heating demand; conversely, in the heat storage mode, the valve is closed to prevent hot water from flowing out, thereby achieving thermal energy storage.

[0035] Pumps are responsible for circulating water in the thermal storage system to ensure that heat is efficiently transferred between the thermal storage medium and the user. The pumps are generally powered by electricity generated by the photovoltaic system, ensuring that the system operates efficiently when sunlight conditions are good. The operation of the pumps is essential to maintaining the dynamic balance of the system and helps achieve a fast response.

[0036] The data collection module, prediction module, and control decision module are all deployed on the cloud server platform.

[0037] The electric-thermal coordination method for directional transport of multi-scale electric-thermal energy flows described in the present invention solves the technical problem of solving the incoordination between the heat storage system and the photovoltaic system in the process of energy utilization and scheduling. The present invention realizes efficient coordination between the heat storage system and the photovoltaic system, can monitor and predict the demand for electric energy and thermal energy in real time, dynamically adjust the heat storage and release modes according to historical and real-time data, thereby optimizing energy utilization efficiency, and adopts machine learning models to improve the accuracy of power generation and load prediction, and enhance the system's responsiveness to environmental changes. It not only reduces energy waste, but also effectively improves the utilization rate of renewable energy, thereby facilitating intelligent energy management.

Claims

1. An electrothermal synergistic method for directional transport of multi-scale electrothermal energy flow, characterized in that: The steps include: Step 1: The data collection module obtains the water storage temperature data Tstorage, the outlet water temperature data Tout and the return water temperature data Treturn of the thermal storage system in real time through the Internet, constructs a thermal storage temperature data set, and simultaneously obtains the voltage Vthermal and current Ithermal of the thermal storage system power generation output, calculates the power generation Pthermal of the thermal storage system; constructs a thermal storage power generation data set; The data collection module obtains the voltage Vpv and current Ipv output by the photovoltaic system in real time through the Internet, calculates the power generation Ppv of the photovoltaic system, and generates a photovoltaic power data set; Step 2: The prediction module retrieves the photovoltaic power data set and predicts the photovoltaic power generation Ppv,forecast of the photovoltaic system in the next time period t through regression analysis; Retrieve the thermal storage power generation data set and use the decision tree regression model to predict the thermal storage power generation Pthermal,forecast in the next time period t; Step 3: The forecast module calculates the total forecast power generation Ptotal,forecast; Ptotal,forecast = Ppv,forecast + Pthermal,forecast; Step 4: The control decision module obtains historical heating demand data and historical electricity consumption data, draws a time series diagram of historical electricity consumption based on the historical electricity consumption data, and calculates the predicted user load Pload,next for the next time period t based on the ARIMA model; Retrieve the total predicted power generation Ptotal,forecast calculated by the prediction module, make decisions, and perform optimization: decision making: If Ptotal,forecast < Pload,next, the thermal storage system enters the heat release mode; If Ptotal,forecast ≥ Pload,next + s, the thermal storage system enters the thermal storage mode; Optimizing decisions: If Tout < Tdemand, the thermal storage system gives priority to heating; If Tstorage > Tmax, the thermal storage system considers entering the heat release mode; Tdemand is the target temperature for user heating demand; Tmax is the maximum safe temperature of water stored in the thermal storage system; s is the preset safety margin; According to the load fluctuation frequency Fload, the power generation change frequency Fpv, the system response time Tresponse and the heating demand change Fheat, the mode switching duration Tswitch of the thermal storage system is calculated according to the following formula: T switch =k × [ 1 / (F load + F pv + F heat )] + T response ； Among them, k is the adjustment coefficient, which is adjusted according to the actual operation of the thermal storage system and reflects the influence of various factors on the duration; Step 5: The control decision module sends the control decision to the thermal storage system. The thermal storage system starts the corresponding valves and pumps according to the control decision, adjusts the coordinated exchange of thermal energy and electrical energy, and realizes dynamic regulation of energy.

2. The electrothermal synergistic method for directional transport of multi-scale electrothermal energy flow according to claim 1, characterized in that: When executing step 1, the power generation Pthermal of the thermal storage system is calculated by the following formula: Pthermal = Vthermal × Ithermal; The power generation Ppv of the photovoltaic system is calculated by the following formula: Ppv = Vpv × Ipv.

3. The electrothermal synergistic method for directional transport of multi-scale electrothermal energy flow according to claim 2, characterized in that: When executing step 2, the specific calculation formula of photovoltaic power generation Ppv,forecast is as follows: P pv,forecast = a 0 + a 1 × P pv,real + a 2 × P pv,hist + a 3 × Weather Factors + ϵ ; Among them, Ppv,forecas is the predicted photovoltaic power generation; a0 is the intercept, which represents the power generation when all independent variables are 0; a1, a2 and a3 are regression coefficients, which respectively represent the influence of real-time photovoltaic power generation, historical power consumption and weather factors on power generation; ϵ is the error term, which represents the prediction error of the model; Ppv,real is the real-time photovoltaic system power generation, Ppv,hist is the historical photovoltaic power generation data, and Ppv,real and Ppv,hist are calculated using the data in the photovoltaic power data set; Weather Factors is weather data, including solar radiation intensity; The specific calculation formula of thermal storage power generation Pthermal,forecast is as follows: For a given input feature vector X, use the trained decision tree model to make predictions: X = [Tstorage,Tout,Treturn,Pthermal,past]; Pthermal,forecast = Tree(X); Pthermal,past refers to the power generation data of the thermal storage system in a certain period of time in the past, which is obtained through the thermal storage power generation data set; Tree represents the decision tree model.

4. The electrothermal synergistic method for directional transport of multi-scale electrothermal energy flow according to claim 1, characterized in that: When executing step 4, the historical electricity consumption data is the historical user electricity consumption data obtained by the control decision module from the power grid system through the Internet; The execution steps of the control decision module are as follows: Step 4-1: The control decision module obtains historical power consumption data; Step 4-2: The control decision module cleans the historical power consumption data, removes outliers and missing values, constructs a historical power consumption data set, and arranges the data in the historical power consumption data set in time series, specifically in hourly, daily or weekly time series; Step 4-3: Draw a time series graph to obtain the trend, seasonality and cyclical characteristics of the data; Draw the autocorrelation graph ACF and partial autocorrelation graph PACF; Step 4-4: Determine the parameters of the ARIMA model, including the number of autoregressive terms p1 , the number of differences d, the number of moving average terms q, the number of autoregressive terms p is the number of significant lags obtained from the PACF plot, the number of moving average terms q is the number of significant lags obtained from the ACF plot, and the number of differences d is the number of times the difference is performed when the unit root test is used to check the stationarity of the time series; Step 4-5: Build ARIMA model: ARIMA(p,d,q); Step 4-6: Fit the ARIMA model using historical electricity consumption data; Step 4-7: Use the fitted ARIMA model to predict the next time period. The prediction formula is: P ＇ load,t = μ + ϕ 1 P load,t−1 + ϕ 2 P load,t−2 + ...+ ϕ p1 P load,t−p1 + θ 1 ϵ t−1 + θ 2 ϵ t−2 +...+ θ q ϵ t−q ； in, P ＇ load,t is the predicted user load; μ is the constant term of the model; ϕ is the autoregressive coefficient; θ is the moving average coefficient; ϵ is the error term; t Indicates a time period; q is the number of moving average terms; p1 is the number of autoregressive terms; load Indicates the user load side; Step 4-8: Use the predicted value as the predicted user load Pload,next for the next time period; Step 4-9: Use the mean square error (MSE) or root mean square error (RMSE) indicator to evaluate the predictive ability of the ARIMA model; Step 4-10: Regularly adjust the ARIMA model parameters based on the difference between actual electricity consumption and the predicted value.

5. The electrothermal synergistic method for directional transport of multi-scale electrothermal energy flow according to claim 1, characterized in that: The load fluctuation frequency Fload is obtained by analyzing the changes in historical power consumption data and calculating the frequency of load changes per unit time; The power generation change frequency Fpv is obtained by monitoring the power generation of the photovoltaic system and recording the number of changes in power generation per unit time; The system response time Tresponse is determined by the system's reaction time to changes in the input signal and is a preset fixed value; The heating demand change Fheat is obtained by analyzing the changes in historical heating demand data, which represents the number of heating demand changes within a statistical unit time.

Images